The European Artificial Intelligence Act, Regulation (EU) 2024/1689, has been applying in stages since February 2025, and it was amended on 27 July 2026 by Regulation (EU) 2026/1744, the Digital Omnibus.
Two sentences to place yourself. The Article 50 transparency obligations have applied since 2 August 2026. The high-risk obligations were postponed to 2 December 2027 for Annex III and 2 August 2028 for Annex I.
Two questions decide the rest, in this order. What is your role, system by system: provider, deployer, or both on different systems. Then what risk tier each system falls into. The two answers cross, and that crossing gives you the obligations and their dates.
The two silos are listed at the foot of this page. The pages below answer a narrower question directly.
This site explains the European regulation on AI: what it says, who it applies to, and when each obligation takes effect.
It follows changes to the text and the positions taken by the authorities. Every page carries its update date, and every dated finding says when it was checked.
Every statement cites the Regulation, article and paragraph, word for word. Anything that comes from a European Commission interpretation is flagged as such: guidelines do not have the force of the text.
The questions organisations actually ask
The timeline: every deadline that binds an organisation, from February 2025 to December 2030, each with the provision it rests on.
What the Digital Omnibus changed: what Regulation (EU) 2026/1744 actually moved, and what it added.
Is my system high risk?: the two routes in, the eight areas of Annex III, and the Article 6(3) exit filter that almost nobody reads correctly.
Is what I am doing prohibited?: the ten bans in Article 5, and the far longer list of things that are widely believed to be banned and are not.
What must I disclose, mark or label?: Article 50, addressee by addressee, and the 2 December 2026 catch-up deadline.
What does AI literacy actually require?: the one obligation that falls on every provider and every deployer, whatever the risk level.
If you build on a model
General-purpose AI models: Chapter V obligations, the systemic-risk threshold, the point at which fine-tuning makes you a provider, and the enforcement powers the Commission holds over this chapter alone.
On proving it. As at 24 August 2026, the date of our last check, no harmonised standard reference had been published in the Official Journal of the European Union under Regulation (EU) 2024/1689, so the presumption of conformity everyone invokes is available to nobody. What does and does not count as evidence is set out in the standards silo.
How this site works
Every factual statement here is anchored in the text of the Regulation in its consolidated version of 27 July 2026, with the provision it rests on, on the understanding that the authentic text is the one published in the Official Journal, the consolidated version having documentary value only.
Where a document is still a draft, which is the case for the guidelines on high-risk classification and on Article 50, the page says that too.
Quotations are reproduced from the official English texts and are never translated from another language version. Where no English version exists, the page reports the substance and says why it is not quoting.
This is an independent resource. It is not legal advice.
Sources. Regulation (EU) 2024/1689, consolidated text of 27 July 2026 and Regulation (EU) 2026/1744, both consulted on 27 August 2026.
Frequently asked questions
Does the EU AI Act already apply, or has it been postponed?
Both, depending on which part of the text, and that is the most widespread confusion on the subject. Applying today: the prohibited practices and AI literacy since 2 February 2025, the general-purpose AI model obligations since 2 August 2025, the penalty regime since the same date with the exception of Article 101, which falls under the general date of 2 August 2026, and the Article 50 transparency obligations since 2 August 2026. Regulation (EU) 2026/1744 moved the high-risk dates to 2 December 2027 for Annex III and 2 August 2028 for Annex I, and it added new ones, starting with 2 December 2026 for two of the prohibitions. The detail deadline by deadline, each with the provision it rests on, is on the timeline page.
Is the 2024 text still the one to read?
Not on its own. Regulation (EU) 2024/1689 was amended on 27 July 2026 by Regulation (EU) 2026/1744, the Digital Omnibus, which rewrote a number of articles and inserted new ones, among them Article 4a on the processing of special categories of personal data for bias detection and correction. Reading the July 2024 version without its amendments leads to errors of date and of substance alike. This site works from the consolidated version of 27 July 2026, on the understanding that the authentic text remains the one published in the Official Journal.
Does this site give legal advice?
No. It is a documentary resource: we read the official text, we quote what it says word for word, and we date every check. Nothing published here replaces a lawyer's opinion on a particular situation.
This content is a technical and regulatory information resource. It does not constitute legal advice.